The association of microtubules with tight junctions is promoted by cingulin phosphorylation by AMPK

The structure of epithelial cell sheets, in which cell–cell adhesion is highly organized, is critically dependent on the association of cytoskeletal components with apical cell–cell adhering junctions (including tight junctions [TJs] and adherens junctions [AJs] and desmosomes; Gumbiner, 2000; Tsukita et al., 2001; Perez-Moreno et al., 2003; Franke, 2009; Meng and Takeichi, 2009). Therefore, the dynamic interactions of actin and 10-nm filaments with the plasma membrane at the cell–cell junctions have been well studied in the respect of their roles in organizing cell–cell adhesion/cytoskeleton and in transducing inter- and intracellular signaling (Bornslaeger et al., 1996; Etienne-Manneville and Hall, 2002; Sugimoto et al., 2008; Baum and Georgiou, 2011).

Microtubules (MTs) that interact with plasma membranes are reported to bind MT plus end–tracking proteins at the scaffold in the AJs with their plus ends or to bind Nezha/calmodulin-regulated spectrin-associated proteins and ninein in the AJs with their minus ends (Moss et al., 2007; Shaw et al., 2007; Meng et al., 2008; Meng and Takeichi, 2009). How MTs interact with cell–cell adhering junctions provides clues to how the dynamic arrangements of MTs are regulated in cells. Further analyses of this system should shed light on the molecular bases of the cell–cell junction-based organization of microtubular networks.

Cellular MTs form two types of networks, those composed of centrosomal MTs and those composed of noncentrosomal ones, and the balance between them is thought to be regulated by cell type–dependent cues (Bacallao et al., 1989; Reinsch and Karsenti, 1994; Bartolini and Gundersen, 2006). In epithelial cells, unlike many other cell types, such as fibroblasts, the noncentrosomal MTs dominate; they are oriented apicobasally, although the dynamics of their arrangements have not been well analyzed. Thus, epithelial cell–specific cues likely play a role in their unique MT arrangements. In addition, cell–cell adhesions between epithelial cells are highly organized, particularly in epithelial cell sheets, and the unusual arrangement of MTs may be related to the functions of cell–cell adhering junctions.

A potentially fruitful approach to understanding the relationship between the cell–cell adhesion system and MTs’ organization in epithelial cell sheets would be to examine the effects of altering cell–cell adhesion system on MT organization. Here, we examined epithelial cell sheets using structured illumination microscopy (SIM) and found a new noncentrosomal MT network, which was organized into a planar apical structures. Furthermore, in addition to associating end-on with the TJs, the MTs were aligned laterally to TJs, with the side of the filaments apparently at the site of the MT–TJ association. We found that the interaction between the MTs and TJs was mediated by cingulin, through its AMP-activated protein kinase (AMPK)–dependent phosphorylation. These results point to the role of the TJ as an organizing site for the apical MT network’s formation. When the association of MTs with TJs was perturbed by cingulin knockdown (KD), by expressing dephosphomimetic mutants of cingulin, or by an AMPK inhibitor, the morphogenesis of the cells’ 3D colonies was markedly compromised. These findings reveal new information about the distribution and function of the planar apical networks (PANs) of MTs in epithelial cell sheets.

Here, we immunostained polarized cell sheets, formed by the Eph4 epithelial cell line, which are derived from the mouse mammary gland, for α-tubulin and ZO-1 (a TJ marker), and observed them by SIM. The results revealed a PAN of noncentrosomal MTs (PAN-MTs), just beneath the apical plasma membrane, at the same level as where the TJs are located (Figs. 1 A and S1 A and Video 1). (In contrast, most of the other noncentrosomal MTs remained aligned in the apicobasal direction.) These PAN-MTs could not be clearly identified by conventional immunofluorescence microscopy, which may explain why it was overlooked previously (Fig. 1 B).

Notably, soon after cell–cell adhesion was established, the PAN-MTs appeared as a separate network from the centrosomal MTs in the apicobasal view (Figs. 1 A and S1 A and Video 1). In contrast, long after cell–cell adhesion was established, centrosomes were located in the PAN-MT region, but they were no longer associated with MTs (Fig. S1 A and Video 2). Thus, the PAN-MTs form a noncentrosomal MT network that has not been previously described. In addition, we found the edges of the PAN-MTs associated with the cell–cell junction in a side-by-side fashion (Fig. 1 C). Next, to trace the ends of the PAN-MTs, we immunostained for α-tubulin, for EB1 as a plus-end marker of MTs and for Nezha as a minus-end marker of MTs. The minus and plus ends of MTs coexisted in the apical regions without any connections to centrosomes (Fig. S1 B and Video 3). Hence, the planar MTs are most likely noncentrosomal because they did not colocalize with centrosomes. This point remains to be further clarified in a future study.

To evaluate the MT–TJ interaction, we performed a gel overlay assay of MTs (stabilized in their polymerized form by taxol) on polyvinylidene difluoride (PVDF) membranes, on which were blotted polypeptides from extracts of the epithelial cell–cell junction fraction isolated from liver (Tsukita and Tsukita, 1989; Furuse et al., 1993); this fraction contains a substantial amount of TJs. As shown in Fig. 1 D, the MTs showed strong binding to a 140-kD polypeptide (J-MAP 3, which was identified as cingulin by direct peptide sequencing) and weaker binding to three other bands (J-MAP 1, 2, and 4). We next asked whether cingulin mediated the MT–TJ interaction. In coprecipitation assays, α-tubulin was pulled down by anti-HA antibodies from HA-cingulin–overexpressing HEK293 cells, and an anti–α-tubulin antibody pulled down HA-cingulin (Fig. 2 A). Therefore, we identified cingulin as a MT-binding protein.

To examine the interaction between cingulin and MTs in more detail, we performed a domain analysis, in which we divided cingulin into three domains, a head domain (1–333 aa) and two rod domains, rod 1 (334–760 aa) and rod 2 (761–1,193 aa). The head domain of cingulin was previously reported to associate with actin, ZO-1, and ZO-2. On the other hand, two rod domains are coiled-coil regions that are involved in dimer formation (Citi et al., 2000; D’Atri et al., 2002). To examine the binding affinity of each domain to endogenous α-tubulin, we overexpressed the HA-tagged construct of full-length cingulin, or of the separate head, rod 1, or rod 2 domain, in HEK293 cells. The full-length and head domain of cingulin, but not the rod 1 or rod 2 domain, bound to α-tubulin, indicating that cingulin binds to MTs through its head domain (Fig. 2 B). It seemed that α-tubulin interacted better with the cingulin head domain than with the full length of cingulin, suggesting some conformational regulation of the binding between α-tubulin and cingulin in its full length, which was related to the phosphorylation of head domain of cingulin, as shown in Figs. 3 C and S3 B.

Furthermore, when the head domain of cingulin was divided into the subdomains of 1–202 aa and 203–333 aa, respectively, α-tubulin bound to the 1–202-aa sequence and ZO-1 to the 203–333-aa sequence, suggesting that the bindings of α-tubulin and ZO-1 to cingulin are not mutually exclusive (Fig. S1 C). Finally, we confirmed the binding between the proteins by using an endogenous coimmunoprecipitation assay; α-tubulin was pulled down by the anti-cingulin antibody, and an anti–α-tubulin antibody pulled down endogenous cingulin (Fig. 2 C).

We next asked whether cingulin mediated the side-by-side association of MTs with TJs. For this analysis, we generated cingulin KD Eph4 cells by the stable transfection of KD vectors (Fig. 2 D). Suppression of cingulin mRNA has no effect on AJ and TJ protein expression (Fig. S2 A), although immunofluorescence microscopy showed that the suppression of cingulin expression markedly decreased the side-by-side lateral association of MTs with TJs (Fig. 2 E). To exclude the possibility that the observed disruption was caused by a side effect of cingulin KD, RNAi-resistant cingulin was transfected into cingulin KD cells, which restored the MT–TJ association. In addition, the MT–TJ association was disrupted in ZO-1 knockout Eph4 cells, in which cingulin is known to be dissociated from TJs (Fig. S1 D; Umeda et al., 2004). These findings collectively indicated that cingulin plays a major role in the side-by-side association of MTs with TJs. To examine the dynamics of the PAN-MTs, we transfected RFP-EB1 into Eph4 and cingulin KD cells, to trace the EB1 signals as the plus-end marker of MTs. In Eph4 cells, the EB1 signals were located parallel to the TJs. On the other hand, in cingulin KD cells, EB1 signals tended to be located end on with respect to the membranes at points of cell–cell adhesion (Videos 4 and 5).

Cingulin is also reported to associate with actin filaments (D’Atri and Citi, 2001) as well as with guanine nucleotide exchange factor (GEF)–H1 and p114 RhoGEF, as shown in MDCK and Caco-2 cells, respectively (Aijaz et al., 2005; Terry et al., 2011). There was no difference in actin filament arrangement, myosin light chain phosphorylation, p114 RhoGEF, or GEF-H1 between wild-type Eph4 and cingulin KD Eph4 cells (Fig. S2, B–E). We also did not detect differences in Rho activity, as shown in fluorescence resonance energy transfer (FRET) analyses, between the wild-type and cingulin KD cells (Videos 6 and 7). These results collectively indicated that cingulin mediates the lateral association of MTs with TJs, in a manner that does not involve Rho-related signaling.

We next examined the mechanism regulating cingulin’s association with TJs. Cingulin is phosphorylated on its serine residues, similar to other TJ proteins, such as occludin and JAM-A (Citi and Denisenko, 1995; Seth et al., 2007; Raleigh et al., 2011; Iden et al., 2012). Cingulin has two AMPK target motifs L/SXXRXS/T at its serine-132 and -150 residues (Fig. 3 A), and TJ assembly is reported to be facilitated by the AMPK activator AICAR (5-aminoimidazole-4-carboxamide ribonucleotide; Zhang et al., 2006; Zheng and Cantley, 2007). We therefore examined whether cingulin is a substrate of AMPK. We first analyzed the binding of AMPK to cingulin, by coimmunoprecipitation experiments with exogenous HA-cingulin and V5-AMPKα1 expressed in HEK293 cells. The results showed that both proteins were coimmunoprecipitated by an anti-HA antibody, indicating that they bound each other (Fig. 3 B).

Next, to examine whether cingulin was a substrate of AMPK, we generated dephosphomimetic mutants of GST-cingulin, consisting of single (S132A or S150A) and double (S132A/S150A) dephosphomimetic mutants of cingulin fused to GST. GST-cingulin (wild type) and its dephosphomimetic mutants were purified and incubated with GST-AMPK (α1/β1/γ1) in the presence of ATP and AMP. The phosphorylation signals in the GST-cingulins were then examined using Pro-Q diamond, which detects phosphorylated proteins. Signals were detected in the bands of GST–wild-type cingulin, weaker signals were detected in the single mutant of S132A or S150A, and almost no signal was detected in the double dephosphomimetic mutant S132A/S150A (Fig. 3 C). Thus, cingulin is probably a phosphorylation substrate of AMPK, and S132 and S150 are AMPK’s target sites.

We then examined the effects of the AMPK inhibitor compound C on cingulin’s association with MTs in Eph4 cells. Immunofluorescence microscopy showed that the AMPK inhibitor affected the association of MTs with TJs, much as observed in cingulin KD cells, but not the localization of cingulin (Fig. 3 D). These results suggested that cingulin’s function in mediating the MT–TJ association was regulated by its phosphorylation by AMPK. To further define the role of cingulin in the formation of the planar MT network, we examined calcium-switched formation of TJs. Because KD of cingulin and AMPK inhibitor induced detachment of the PAN-MTs from TJs, but did not affect the number of MTs in the apical network, it was likely that cingulin contributed to the stabilization of the MT–TJ interaction but not to the formation of the apical network of MTs (Fig. S3 A).

We addressed whether AMPK-mediated phosphorylation regulated cingulin’s binding to MTs. For this purpose, lysates prepared from transfectants of HA-tagged wild-type cingulin or its dephosphomimetic mutants (S132A, S150A, and/or S132A/S150A) were immunoprecipitated with anti–α-tubulin. HA signals were detected in the wild-type cingulin bands, weaker signals were detected in the cingulin S132A or S150A bands, and almost no signal was detected in the double dephosphomimetic mutant S132A/S150A bands (Fig. 4 A). These findings supported the idea that the AMPK-mediated phosphorylation of cingulin regulated its binding to α-tubulin. Because compound C did not decrease the binding of α-tubulin with the head domain of cingulin, it was most likely that AMPK phosphorylation induced some conformational changes in cingulin to expose its binding sites to α-tubulin. Further studies are required to confirm this point (Fig. S3 B).

Next, we examined whether the AMPK-mediated phosphorylation of cingulin regulated the lateral interaction of MTs with TJs. The single or double phosphorylation site mutants localized to TJs but could not rescue the defective MT–TJ arrangement caused by cingulin KD (Fig. 4 B), and the double phosphomimetic mutant S132D/S150D rescued the MT–TJ arrangement caused by cingulin KD and inhibition of AMPK (Fig. S3 C). Taken with the finding that AMPK-mediated phosphorylation was the major phosphorylation in cingulin, it appears to play a critical role in cingulin’s association with MTs, which is the basis of the interaction of MTs with TJs.

Finally, we examined the biological relevance of the MT–TJ association in epithelial cells. For this analysis, we performed 3D cultures of the following Eph4 cells: wild-type, cingulin KD, cingulin KD revertant expressing RNAi-resistant cingulin, and cingulin KD expressing cingulin dephosphomimetic mutants, in collagen IA gel. When the shape of the colonies was analyzed using ImageJ software, the colonies of wild-type Eph4 cells formed isotropic spheroids without lumen (Figs. 4 C and S3 D). In contrast, the colonies of cingulin KD cells had a distorted, anisotropic shape (Fig. 4 C). The cingulin KD revertant colonies showed the same round shape as the wild-type cells, indicating that the KD of cingulin was the direct cause of the deformation of the 3D Eph4 colonies (Fig. 4 C). Finally, when cingulin dephosphomimetic mutants were expressed in cingulin KD cells, the colonies showed a distorted, anisotropic shape, indicating that phosphorylation of cingulin is critical for the shape of colonies.

We quantified the isotropies of the 3D colonies by representing the colonies as rectangles and determining the isotropic indexes as the ratios of the shortest to the longest lengths. This ratio was significantly different between the 3D colonies of wild-type and cingulin KD cells, 0.83 ± 0.017 (n = 110) and 0.65 ± 0.026 (n = 66), respectively. The ratio in the revertant was 0.78 ± 0.008 (n = 128). Furthermore, branching of the 3D colonies of cingulin KD cells occurred but was not seen in the colonies of wild-type or cingulin KD revertant cells (Fig. 4 D). The expression of phosphomimetic mutants does not significantly show such effects. In addition, Eph4 cells treated with compound C formed the anisotropic colony (0.59 ± 0.012, n = 302; Fig. S3 E). Thus, anisotropy and branching were induced by the absence or dephosphorylation of cingulin. These findings indicated that the AMPK-mediated MT–TJ interaction probably contributes to epithelial morphogenesis, and the apical MT network provides sufficient tension to the apical membrane to form the isotropic spherical shape, pointing to a critical role of the apical configuration of epithelial cell sheets.

In summary, as schematically shown in Fig. 5, we have for the first time revealed a PAN of noncentrosomal MTs (PAN-MTs), which is laterally associated with the TJs through cingulin, in its AMPK-phosphorylated form, by the high-contrast images achieved by SIM. AMPK is a kinase that plays critical roles in the regulation of a wide spectrum of metabolic homeostasis and is reported to generate a variety of biological cues (Leprivier et al., 2013; Miller et al., 2013; O’Neill and Hardie, 2013). This kinase regulates energy-dependent processes in epithelial morphogenesis, cell polarity, and tumor suppression (Lo et al., 2012; Martin-Belmonte and Perez-Moreno, 2012). In this respect, the PAN-MT system is a target of metabolic homeostasis-related AMPK regulation, involved in the apical maturation of epithelial cell sheets and epithelial morphogenesis. These findings increase our basic understanding not only of epithelial cell biology but also of cancer and developmental biology.

Primary antibodies used in this work were mouse anti–α-tubulin mAb (Sigma-Aldrich), rat anti–α-tubulin mAb (Abcam), mouse anti-HA mAb (Covance), rat anti-HA mAb (Roche), and rat anti-GFP mAb (Nacalai Tesque) antibodies. Mouse Anti-V5 mAb (Invitrogen) was gifted by S. Takashima and O. Tsukamoto (Osaka University, Osaka, Japan) and mouse anti-cingulin mAb (antigen: full-length of cingulin) was produced by K. Owaribe (Nagoya University, Nagoya, Japan). Rabbit anti–ZO-1 pAb (antigen: F4 fragment including 30–340 aa; Itoh et al., 1993) and mouse anti-afadin mAb (antigen: full-length of afadin) were generated in our laboratory. Alexa Flour 488–, 568–, and 647–labeled secondary antibodies and rhodamine-conjugated phalloidin were commercially obtained (Invitrogen). HRP-conjugated secondary antibodies were also commercially obtained (BD). Compound C was commercially obtained (EMD Millipore).

To suppress the expression of cingulin in Eph4 cells, oligonucleotides of target sequence were cloned into the H1 promoter-driven RNAi vector (Brummelkamp et al., 2002). The vector was transfected and suppressed the expression of cingulin, and we obtained two clones. The probe sequence was cingulin, 5′-GACCGTTTGTGGTTCTTAAC-3′.

Mouse Eph4 epithelial cells, cingulin KD cells, and HEK293 cells were grown in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum. Transfection was performed using Lipofectamine Plus reagent (Invitrogen) according to the manufacturer’s instructions.

Cells were fixed in cold methanol for 10 min on ice or fixed in 1% formalin for 5 min at RT followed by treatment with 0.1% Triton X-100 in PBS. After blocking for 10 min, cells were incubated with primary antibodies in blocking buffer for 1 h at RT or overnight at 4°C. After washing, cells were incubated with fluorochrome-conjugated secondary antibodies for 1 h at RT. The cells were mounted in fluorescence mounting medium (Dako). The specimens were observed with a photomicroscopy (BX51 and BX70; Olympus) equipped with a 100×, 1.4 NA oil immersion lens, 60×, 1.42 NA oil immersion lens, and 20×, 0.5 NA lens, and with a superresolution SIM (ELYRA S.1; Carl Zeiss) equipped with a Plan Apochromat (100×, 1.46 NA oil immersion lens, 63×, 1.4 NA oil immersion lens, and 40×, 1.4 NA oil immersion lens) with appropriate binning of pixels and exposure time. Photographs were recorded with a cooled charge-coupled device camera (ORCA-ER [Hamamatsu Photonics] or CoolSNAP HQ [Photometrics]). The images were analyzed with MetaMorph (Molecular Devices) or ZEN (Carl Zeiss).

The junctional fraction was prepared from the liver of newly hatched or 2-d-old chicks through the crude membrane and the bile canaliculi (BC) fractions according to the method described previously (Tsukita and Tsukita, 1989). The BC fraction was diluted fivefold (vol/vol) with hypotonic buffer (1 mM NaHCO₃ and 2 µg/ml leupeptin, pH 7.5) and centrifuged at 100,000 g for 30 min at 4°C. The precipitate was dissolved with buffer A (50 mM Hepes, pH 7.5, 1 mM EGTA, 6 M urea, 2 µg/ml leupeptin, and 10 mM APMSF) and centrifuged at 100,000 g for 60 min at 4°C. The resulting supernatant (∼20 mg) was applied to an SP Sepharose column (GE Healthcare). After the column was washed with buffer A containing 50 mM NaCl, the binding proteins were eluted with the same buffer containing 100 mM NaCl and then with buffer A containing 150 mM NaCl. The eluate from the 150 mM NaCl solution was diluted threefold with buffer A and applied to a Q Sepharose column (GE Healthcare). The column was washed with buffer A containing 150 mM NaCl, and bound proteins were then eluted with the same buffer containing 200 mM NaCl. Aliquots of the eluate were subjected to SDS-PAGE (4–7.5% gradient gel) and transferred to the PVDF membrane.

Pig brain tubulin was purified as previously described (Nishida et al., 1987). Purified tubulin (∼1 mg/ml) was polymerized into MTs by incubating for 60 min at 37°C in 3 mM MgCl₂, 1 mM EGTA, 1 mM GTP, 10% DMSO, and 80 mM Pipes, pH 6.8. The sample was then diluted 22-fold in PME buffer (1 mM MgCl₂, 1 mM EGTA, 20 µM taxol, and 80 mM Pipes, pH 6.8) and kept at RT.

The PVDF membrane was blocked with 5% skim milk (Megmilk Snow Brand Co., Ltd.) in PME buffer for 1 h at RT. The membrane was then incubated with 5% skim milk in PME buffer, which contains 45 µg/ml of MTs, for 2 h at 37°C. After washing with PME buffer for 5 min at 37°C three times, the bound polymerized tubulin was detected using an anti–α-tubulin antibody.

HEK293 cells were transfected with expression vectors. Cell lysates were incubated with protein A–Sepharose bound with the anti–α-tubulin or anti-HA antibody. Immune complexes were fully washed and then resuspended in 30 µl SDS sample buffer, and 5- and 20-µl aliquots of each were analyzed by Western blotting.

To prepare total cell lysates for immunoblotting, Eph4 or HEK293 cells were lysed with SDS-PAGE sample buffer, sonicated, and boiled. The protein samples were separated by SDS-PAGE, transferred onto a nitrocellulose or PVDF membrane, and blotted with the appropriate antibodies. For quantification of signals in Western blotting, the densitometric quantification of immunoblot bands with loading control in the same immunoblotting membranes was performed using ImageJ software (National Institutes of Health).

Cingulin phosphorylation assays were performed at 30°C in a reaction volume of 30 µl containing 20 mM Tris-HCl, pH 7.4, 0.3 mM NaCl, 0.2 mM AMP, 0.8 mM MgCl₂, and 0.2 mM ATP, containing 0.1 mM recombinant AMPKα1/β1/γ1 (Carna Biosciences) and either of 1 µg GST-cingulin or GST-cingulin mutants. After 90 min, reactions were terminated by the addition of SDS solution. These samples were separated by SDS-PAGE. The gels were stained with Pro-Q diamond (Invitrogen) according to the manufacturer’s instructions, and the phosphorylation signals were detected by a scanner (Typhoon 9200; GE Healthcare). Densitometric quantification of phosphorylation bands was performed using ImageJ software.

Cells were added to a collagen I (Nitta Gelatin) mixture, gently mixed, and plated onto 12-well transwell insert plates at 5 × 10⁴ cells/well. 3 d after plating, cysts were examined for the immunofluorescence microscopy (Yano et al., 2011). After treatment with collagenase III (Sigma-Aldrich), cells were fixed in cold methanol for 30 min on ice or fixed in 1% formalin for 30 min at RT followed by treatment with 0.1% Triton X-100 in PBS. After blocking for 30 min, cells were incubated with primary antibodies in blocking buffer overnight at 4°C. After washing, cells were incubated with Alexa Flour 488–, 568–, and 647–labeled secondary antibodies for 3 h at RT. Cells were mounted in fluorescence mounting medium (Dako). The specimens were observed with a superresolution SIM (ELYRA S.1) or confocal microscope (LSM 510; Carl Zeiss) equipped with a Plan Apochromat (100×, 1.46 NA oil immersion lens, 63×, 1.4 NA oil immersion lens, and 40×, 1.4 NA oil immersion lens) with appropriate binning of pixels and exposure time. The images were analyzed with ZEN or LSM 510 Meta version 3.0 (Carl Zeiss).

By using ImageJ, an image processing software, we quantified the isotropies of the 3D colonies by representing the colonies as rectangles and determining the isotropic indexes as the ratios of the shortest to the longest lengths.

Data are presented as means ± SE. Whenever necessary, statistical significance of the data was analyzed by performing one-sample t tests. The specific types of tests and the p-values, when applicable, are indicated in the figures.

Fig. S1 shows additional data on the MTs associated with TJs and additional data on the head domain of cingulin. Fig. S2 shows the characterization of cingulin KD cells. Fig. S3 shows the effect of AMPK inhibitor and phosphorylation of head domain of cingulin on MTs arrangements. Video 1 shows the PAN-MTs of Eph4 cells 48 h after being seeded. Video 2 shows the PAN-MTs of Eph4 cells 72 h after being seeded. Video 3 shows the side-by-side association of the PAN-MTs with TJs in an Eph4 cell. Video 4 shows the dynamics of the PAN-MTs in Eph4 cells. Video 5 shows the dynamics in the PAN-MTs of cingulin KD Eph4 cells. Video 6 shows FRET analysis for Raichu-RhoA in the Eph4 cells during 12 and 24 h after Ca²⁺ switch. Video 7 shows FRET analysis for Raichu-RhoA in the cingulin KD Eph4 cells during 12 and 24 h after Ca²⁺ switch.

We appreciate the contribution of Dr. Shoichiro Tsukita, who planned and developed the MT gel overlay assay on purified junctional fractions, together with the authors. We are grateful to Dr. K. Owaribe for the generous gift of the mouse anti-cingulin mAb, to Drs. S. Takashima and O. Tsukamoto for the kind gift of AMPK-related materials, and to Dr. Y. Mimori-Kyosue (Center for Developmental Biology, Kobe, Japan) for the liberal gift of the RFP-tagged EB1 plasmid. We further thank Ms. A. Hagiwara-Yano and Ms. F. Takenaga for technical assistance and members of our laboratories for discussion. We thank graduate students K. Tateishi and R. Tokumasu for schematic drawing and video-imaging materials. We thank Drs. G. Gray, L. Miglietta, and M. Sudol for reading the manuscript.

This work was supported in part by a Grant-in-Aid for Scientific Research on Innovative Areas and for Scientific Research (A) to S. Tsukita from the Ministry of Education, Culture, Sports, Science and Technology, Japan.